private void SaveWeightsBiasAndAgeInPocket(NeuralVector winner, Vector <double> weights, double bias, int age)
{
_pocket.Winner = winner;
_pocket.Weights = weights;
_pocket.Bias = bias;
_pocket.Age = age;
}
private void ValidateVectorLength(NeuralVector vector)
{
if (vector.Data.Count != _numberOfNeurons)
{
throw new InvalidOperationException("Weights or biases dont exist, first teach the network.");
}
}
private bool CalculationsAreCompatibile(double calculationResult, NeuralVector perceptronVector)
{
if (calculationResult >= 0 && perceptronVector.Class == 1)
{
return(true);
}
return(calculationResult < 0 && perceptronVector.Class == -1);
}
public NeuralVector Test(NeuralVector testingVector)
{
ValidateVectorLength(testingVector);
ValidateWeightsAndBiases();
InitializeOutputs(testingVector.Data);
var outputVector = CalculateAllOuputs();
return(new NeuralVector(outputVector));
}
private void FillPatternByOutput(NeuralVector outputVector)
{
var checkBoxes = GetPatternCheckBoxes();
checkBoxes.ForEach(checkBox => checkBox.CheckedChanged -= chBox_CheckedChanged);
for (int index = 0; index < checkBoxes.Count; index++)
{
if (outputVector.Data[index] == 1.0)
{
checkBoxes[index].CheckState = CheckState.Indeterminate;
}
else
{
checkBoxes[index].CheckState = CheckState.Unchecked;
}
}
checkBoxes.ForEach(checkBox => checkBox.CheckedChanged += chBox_CheckedChanged);
}
public static Pattern MapToPattern(NeuralVector vector, int lineCount, int columnCount)
{
var pattern = new Pattern
{
Lines = new Line[lineCount]
};
for (int l = 0; l < lineCount; l++)
{
pattern.Lines[l] = new Line
{
Columns = new double[columnCount]
};
for (int c = 0; c < columnCount; c++)
{
pattern.Lines[l].Columns[c] = vector.Data[l * columnCount + c];
}
}
return(pattern);
}
private double CalculatePerceptronFunction(NeuralVector perceptronVector, Vector <double> weights, double bias)
{
return(perceptronVector.Data * weights + bias);
}
private bool ClassIsCorrectlyClassified(NeuralVector perceptronVector, int classifiedClass)
{
return(perceptronVector.Class == classifiedClass);
}
private void AdjustBias(ref double bias, NeuralVector perceptronVector)
{
bias += perceptronVector.Class;
}
private void AdjustWeights(ref Vector <double> weights, NeuralVector perceptronVector, double learningRate)
{
weights += learningRate * perceptronVector.Class * perceptronVector.Data;
}
